Informatics 43 – April 28, 2015

Fun with Models Fashion Student Model = Ideal

Fun with Models Fashion Student Model = Ideal

Fun with Models Software Process Model Waterfall Prototype Spiral Architecture Model Layered Client Server Model View Controller Model = Simplified Representation

Fun with Models public class Elephant { int weight; // in kg String name; Species species; public void forage() {.... } Model = Simplified Representation

How to choose a Software Process Model? A Software Process Model can be thought of “prescriptively” – providing rules to follow in the future or “descriptively” – serving to describe or characterize what has happened in the past

The Waterfall Software Process Model

The Prototyping Software Process Model

Incremental Software Process Model

Spiral Model

Risk-driven, focus on considering alternatives Each loop through the cycle: Identify a high-risk sub-problem or aspect Resolve the risk (as far as possible) A software project goes through the loop many times Every person on the development team does not have to be at the same stage of the loop

Software Risks Some of Boehm’s Top 10 1.Personnel shortfalls 2.Unrealistic schedules and budgets 3.Developing the wrong software functions 4.Continuing stream of software changes 5.Shortfalls in externally furnished components 6.Shortfalls in externally performed tasks

Agile Software Development We will come back to this topic later in the quarter…

Quality Assurance What software qualities do we want to assure? Correctness

Quality Assurance What software qualities do we want to assure? Correctness How? Testing Inspection and review the source code Proofs, formal methods Static analysis (a program inspects the source code)

Testing Challenges Edsger Dijkstra: “Testing can show the presence of bugs, but not their absence.” True? Why?

Testing Challenges Edsger Dijkstra: “Testing can show the presence of bugs, but not their absence.” True! Because… Software systems are discrete, not continuous. Too many input-output pairs (test-cases) to exhaustively verify, so we must take a small sample. Example: multiplier, inputs two integers, outputs one integer

Testing Challenges Let’s be practical. Testing can only select a very small set of inputs. Our goal should be to choose the best ones. What are the best 5 test cases for a multiplier? (In other words: what 5 test cases, if they don’t reveal any bugs, will give you the most confidence that the multiplier is working correctly?)

Result of running test cases Confidence in module being tested Number of test cases with correct outputs % Sweet spot?

The landscape of “bugs” Error (in programmer’s mind) Fault or defect (in code) Failure (in execution or output)

The landscape of “bugs” Error (in programmer’s mind) Fault or defect (in code) Failure (in execution or output)